The seven-tier ladder: from compression at 1.50 to chaos at 2.50.

What the ladder prices

One question per rung: how far can noise reach before it means something?

An ATR trail's entire job is distinguishing retracement from reversal, and the coefficient is where that judgment gets quantified: trail distance is ATR times coefficient, so the coefficient is a statement about how many units of current volatility constitute ordinary noise. In compression, noise is small and a 1.50 multiple already clears it — anything further is probably real. In extreme conditions, candles routinely travel distances that would be reversals in a quiet market, and a trail priced for quiet gets executed by noise; 2.50 buys the runner survival room. The ladder is that judgment pre-made for seven states, so the live decision reduces to identifying the state.

The rungs

Seven states, and why the top got the resolution.

The redesign's core insight was distributional: the five-tier model spent its resolution evenly while the decision difficulty concentrates at the top. Below expansion, coefficient differences are modest — 1.50 versus 1.60 rarely decides a trade. Above it, the states diverge fast: Strong Expansion at 1.85, Outer Band Ride at 2.00 (a genuinely distinct state — price riding the band under sustained pressure, the persistence input's signature), Very High Expansion at 2.20, and Extreme at 2.50. Those four upper rungs replaced what the old model treated as roughly two, and they're exactly where the trend branches' fat-tail capture gets decided — the big continuation days are, almost by definition, upper-ladder days, and a ladder that can't tell 1.85 conditions from 2.50 conditions mismanages the trades that matter most.

The scope rule

The ladder sets trail width — it never picks the branch.

The transcript locked this interpretation in explicitly, and it's the most commonly violated boundary in volatility tooling: the coefficient is not for deciding whether to trade Normal or Trend. The ATR/BE regime switch answers Normal-versus-Trend; the ladder answers how much breathing room the ATR trail deserves once trend management is already in use. Collapsing the two turns a high tier into a trend signal — 'volatility is expanding, so this must be a trend trade' — which is precisely the regime-forcing failure the launch checklist polices. The ladder activates after the branch decision, inside it, on the single parameter it owns. High volatility with a Normal-regime read is still a Normal trade; it just isn't the ladder's business.

• Sequence: regime switch picks the branch → branch doctrine authorizes a trail → the ladder prices the trail. Never backwards.
• A T6 read is not a trade invitation — extreme volatility is as often a reason for the discretion layer to decline as to engage.
• Normal-branch trades don't consult the ladder at all: static 2R management has no trail to price.

Living with the ladder

Fixed rungs, evidence-driven assignment, documented changes.

The ladder's values are production rules, and the maintenance doctrine treats them accordingly: do not change coefficient ladders casually, and every change goes through the Version Log. What varies live is only the assignment — which rung today's evidence supports, read through the engine mode's strictness — never the rungs themselves. This is the same architecture as the risk tiers: a fixed structure whose selection is dynamic, because a ladder that bends per trade isn't a ladder, it's a mood with decimals. When accumulating audit-log evidence genuinely suggests a rung is mispriced — coefficient-versus-outcome reviews showing systematic giveback at T5, say — that's a calibration finding, and it travels the documented-change path like every other production adjustment.

The key idea

The ladder converts a feeling into an address.

'It's volatile out there' is the kind of observation that used to justify anything. The ladder replaces it with an address: T4, Strong Expansion, 1.85 — assigned from entered evidence, adjusted for timeframe authority, comparable across every trade that ever logged one. The runner gets breathing room proportional to the actual weather, the audit log gets a decision it can grade later, and the operator gets the thing MARS keeps manufacturing at every layer: a defensible answer where a vibe used to be.